On Fri, Jun 13, 2025 at 08:52:55AM +0100, Richard Sandiford wrote:
> > 2025-06-12 Jakub Jelinek <ja...@redhat.com>
> >
> > * cfgexpand.cc (construct_init_block): If first_block isn't BB_RTL,
> > has any PHI nodes and false_edge->dest_idx before redirection is
> > different from make_single_succ_edge result's dest_idx, swap the
> > latter with the former last pred edge and their dest_idx members.
>
> OK, thanks. But now that there are two instances, I wonder if it would
> be worth hiding this detail in a helper function?
Will try to come up with something incrementally if that is ok.
> It looks like the remaining direct use of redirect_edge_succ and
> make_single_succ_edge is in construct_exit_block, but I assume that
> wouldn't be a problem because the successor is always the exit block.
Yes, plus it is also fine because this is after ranger has been disabled and
all bbs converted to BB_RTL. So there are no PHI nodes on any bb at that
point.
Jakub
